Google Webmaster? Google Gives Site Owners an AI Button, But Not a New Deal

Brett Tabkeon

Google’s new Search generative AI control is useful, overdue, and beyond limited – to the point of uselessness? Site owners should treat it less like a gift and more public smoke-n-mirrors obfuscation behind a new lever inside a black box they still do not control.

Google has Google has a new Search Console setting that lets some site owners manage whether their links and content can appear in Google’s generative AI Search features. That includes AI Overviews, AI Mode, and generative AI features in GDiscover. In plain English: publishers can choose whether Google’s AI answers are allowed to show their content, link to their pages, and use their pages to help ground those AI responses. WTAF!?

That sounds like control, but filtered through Google’s Trumpian Reality Distortion Field? It is, at least compared with the old bargain, where a site owner’s practical choices were mostly “allow Googlebot” or GFY option to “vanish from Google Search.” But SEOs and publishers have every reason to read the fine print slowly and translate from the Google Corporate speak reality distortion field. This is not a full reset of the relationship between Google and the web. This is Google throwing a bone that contains no meat, no marrow, or anything remotely useful to address the current reality of the piss off web.

What Google is sorta offering:

  • Include your site in generative AI Search features, which is the default setting.
  • Exclude your site from generative AI Search features, which means your links and content should not appear in those AI answers.
  • Inherit the setting from a parent Search Console property.

The uncomfortable part: opting out also opts you out of visibility

The most important sentence in Google’s documentation is not the one about control. It is the one about consequence. If a site excludes itself, Google says it will not receive traffic or impressions from those generative AI features. That may sound obvious, but it frames the decision as a trade: protect your content from Google’s AI answer layer, or remain eligible for whatever traffic that answer layer might still send.

For large publishers, that may become a negotiation question. For smaller site owners, it is more smelly, cheesy, and awkward. If AI Overviews and AI Mode continue to take up more of the search system, then opting out could mean stepping away from Google visibility. Staying in could mean helping power answers that satisfy the user before the click (dammed if you do – dammed if you don’t). That is not even a choice. It is the appearance of choice.

“Remember, the early bird gets the worm, but the second mouse gets the cheese.” -anon

This does not solve the AI training question

Google is also careful to separate Search generative AI controls from AI training controls. The Search Console toggle is about appearing in Search generative AI features. It is not the same as blocking content from being used to train Google’s models. For that, Google points site owners to Google-Extended, the robots.txt token used for certain Gemini and Vertex AI training and grounding controls.

That distinction matters. A publisher may reasonably ask: “Am I controlling how my content appears in AI answers, how it trains AI systems, or both?” Google’s answer is: those are separate controls. SEOs will need to treat them separately too. Search Console settings, robots.txt, snippet controls, noindex, paywall logic, and licensing policy are now part of the same governance stack. The old “is it indexable?” checklist is too thin for this era.

The skeptical read:Google is not handing site owners the steering wheel. It is adding a dashboard switch after the route has already changed.

The reporting gap is still the sore spot

Google is also rolling out generative AI performance reporting in Search Console. That is welcome, but early coverage indicates that the reporting focuses on impressions, pages, countries, devices, and dates. Click data is the missing tooth. For SEOs, impressions without click behavior can show presence, but not value. It tells you your page was seen somewhere inside the AI layer, but not whether that exposure helped, hurt, substituted for a visit, or merely decorated Google’s answer box.

That makes testing difficult. A publisher can compare AI impressions over time, but the real question is business impact: Did the AI feature send visitors, reduce clicks, change query demand, compress the funnel, or shift attribution into Google’s black box? Without click-level reporting, site owners will still need external tracking, log analysis, rank monitoring, prompt tracking, and old-fashioned revenue math to understand whether participation is worth it.

Why site owners are right to be skeptical

Google’s pitch is that AI Search creates new ways for users to ask deeper questions and discover more sites. That may be true for some queries and some brands. It is also true that answer-first search can reduce the need to click. Early academic work on AI Overviews has already raised concerns about traffic loss, citation behavior, and unsupported claims. Publishers do not need to believe every worst-case scenario to recognize the pattern: Google is moving from ranking pages to synthesizing answers, and that changes who captures the value.

For years, the implied contract was simple enough: websites let Google crawl, Google sends searchers back. AI Search complicates that bargain. If Google uses publisher content to answer the query directly, the publisher may get attribution, an impression, maybe a click, or maybe nothing. That is not the same economic exchange that built the searchable web.

What SEOs should do now

Do not rush to flip the switch sitewide unless there is a clear business reason. Instead, segment the site. Informational explainers, news, affiliate content, product pages, local landing pages, tools, calculators, directories, and gated assets may deserve different AI visibility rules. The best answer will not be ideological. It will be page-type specific.

  • Audit which content is most vulnerable to answer substitution. Simple definitions, commodity FAQs, and generic how-to content are the first places to look.
  • Protect content that has direct commercial value. If a page converts, sells, books, or captures leads, evaluate whether AI exposure helps or cannibalizes that path.
  • Use Google-Extended separately. Do not assume the Search Console AI control handles training use.
  • Build your own AI visibility tracking. Search Console’s AI reporting is a starting point, not the full ledger.
  • Move beyond commodity content. Original data, expert opinion, tools, local proof, reviews, pricing context, and first-party experience are harder for AI systems to flatten into a disposable answer.

The real decision

The question is not simply whether to opt in or opt out. The better question is: where does your site need distribution, where does it need protection, and where does Google’s AI layer turn your work into someone else’s finished answer?

Google’s new control is worth using. It is also worth doubting. The web has entered a phase where crawling, indexing, summarizing, grounding, training, attribution, and traffic are no longer the same conversation. SEOs and site owners should not treat this as a checkbox. Treat it as the beginning of AI search rights management.

The old SEO job was to help pages get found. The new job is harder: help valuable content get found, credited, clicked, and not quietly dissolved into the answer machine.